Theme Seminars: Jonathan Mohr & Ingrid Urberg
Posted on September 3, 2013 by Christopher Thrall
23 September Theme Seminar – Forum Room 2-004, 12:30-2pm Jonathan Mohr – Moore’s Law Meets Amdahl’s Law: Exponential Progress Hits the Wall For the past half-century, progress has been epitomized by Moore’s Law: Gordon Moore’s observation (1965) that the number of components in integrated circuits (ICs) had doubled every year from their invention in 1958, and his …
23 September Theme Seminar – Forum Room 2-004, 12:30-2pm
Jonathan Mohr – Moore’s Law Meets Amdahl’s Law: Exponential Progress Hits the Wall
For the past half-century, progress has been epitomized by Moore’s Law: Gordon Moore’s observation (1965) that the number of components in integrated circuits (ICs) had doubled every year from their invention in 1958, and his prediction that the trend would continue “for at least ten years.” The virtuous circle that reducing the size of the components etched in silicon in an IC allows it to run faster and use less power has enabled the computer technology industry to provide us with exponential growth in the speed and capacity of our computers while keeping costs approximately constant. In the past few years, however, our success at shrinking semiconductor transistors has run up against some physical constraints: we can’t shrink the wires that connect our microprocessors to the circuit boards through which they interact with the rest of the computer system, and we can’t dissipate the heat that is generated by our blazingly fast processors and our massive memory arrays. In order to keep providing us with exponential growth in computer speed and capacity, chip-makers have now been limited to multi-core designs—instead of giving you a smaller and faster processor, you get 2 or 4 or 8 copies of the same processor you bought only one of two years ago. In theory, that should give you an exponential increase in processing power, just as you expected. But Moore’s Law collides with Amdahl’s Law. Gene Amdahl gave us a formula to calculate what degree of speedup we can achieve by executing our code in parallel on multiple processors. It shows that the speedup depends on what proportion of the code to solve a given problem is intrinsically sequential (i.e., can be run on only one processor, while the other processors wait for something to do). For most problems, having two processors will only give a speedup of between 1.2 and 1.5. (It’s a speedup, but it’s not the speedup you thought you were buying.) There is a lesson to be learned from this example: The fact that we made exponential progress in a particular field for 50 years doesn’t mean we can continue to make progress at that rate forever.
Ingrid Urberg – The Nordic Countries as Progressive Role Models: Realities and Myths
The Nordic countries are frequently referred to in both the popular media and in academic circles as being progressive in a myriad of ways ranging from social programs and education to environmental policies and practices. In the past few months alone, the CBC has broadcast programs on elder care in Denmark, education in Finland, Norway’s penal system, and Iceland as “the world’s most feminist country.” In this colloquium talk, I will discuss a few of the ways in which the Nordic countries are held up as models of progress today, but also look at lesser known and frequently overlooked facets of these societies which are arguably regressive and socially conservative.
For more details on Augustana’s Annual Theme, visit http://www.augustana.ualberta.ca/theme/
Alumni, Annual theme, Augustana Campus, Events. | Permalink
Comments are closed.